Autoencoders are unsupervised neural networks designed for image reconstruction, compression, and noise removal. They consist of two parts:
Encoder: Compresses input images into a lower-dimensional latent space representation.
Decoder: Reconstructs the original image from the compressed representation.
Variants of autoencoders include:
Denoising Autoencoders (DAEs): Remove noise by learning to reconstruct clean images from noisy inputs.
Variational Autoencoders (VAEs): Generate new images by learning probabilistic representations of latent space.
Autoencoders are commonly used in medical imaging, facial recognition, and generative models.